The long range goal of this project is to quantitively characterize the epithelium of human urinary bladder from scanned digitized images of stained histologic sections using digital computers, and to develop a data-directed taxonomy for the range of tissues from normal to invasive carcinoma. The PEEP-DECIDE-GRAPH system was used to analyze the digitized images. The object extraction methods met with varying degrees of success, depending on the quality of the tissue sections. A new algorithm for shape and orientation assessment, based on a complex number representation of position along a boundary, was developed and applied.